33 replies to this topic

[Beeker]

What BAD ?

The dm8000 has a old kernel.

[A.A.]

Right but I'd try to avoid that.

As for samba once e2fsprogs headers are in sysroot that's all for building.

 

We can play with PREFERRED_PROVIDER or UPDATE_ALTERNATIVES .

The problem we have here is more at runtime.

 

Cheers

A.A.

[Beeker]

Why so difficult,, What's the problem with the patch?

Edited by Beeker, 2 October 2021 - 16:11.

[WanWizard]

The problem with the patch as it was implemented applied the patch only for the DM8000, but added it to the mipsel feed. Depending on the build order of all mipsel boxes, it was randomly active or not.

 

In an attempt to fix that, the e2fsprogs ARCH was changed to MACHINE_ARCH, but that causes constant rebuilds of all dependent packages.

 

If, like you say, it is safe to apply this patch for all MIPSEL boxes, I'd say that using SRC_URI_mipsel_append would indeed be the simplest fix.

[A.A.]

We have to converge with OE's samba recipe.

Now for hardknott we can still use old samba but in future with honister we'll have to move to recent versions/python3 only.

 

For this simple reason I would try to avoid specific PLi patches.

 

Machine-specific patches for samba are unacceptable imho.

>As long as they don't initializing a HDD, USB stick or SD/CF card, they won't complain.

^ samba does not do that

 

I have never spent much time on PLi's samba, neverthless I don't see any reasons for diverging from OE.

 

 

My 2 cents

A.A.

[A.A.]

I think we should open a new,separate topic about e2fsprogs because I think we all agree samba must stay PACKAGE_ARCH and not MACHINE_ARCH.

[WanWizard]

Yes. But there is no PACKAGE_ARCH defined for samba, and no MACHINE_ARCH dependencies exist. The problem is e2fsprogs that does that, which triggers bitbake to change samba.

 

 

If I look at https://git.yoctopro...e77dc4a36ff047f, I don't see any dependency to DM8000 or MIPSEL at all, it's implemented for all?

[Beeker]

Changes to e2fsprogs triggers almost en entire rebuild I noticed a while ago.

So if you create a e2fspprogrs-dm8000, you have to to that for all packages which depends on e2fsprogs,

 

Openembedded-core has applied the patch for a long time (It's applied ti all boxes in OpenPLI 7.3, and Perhaps OpenPLi 6.x).

 

So I suggest to apply the patch to all mipsel boxes.

[WanWizard]

Agreed, was already busy .

[A.A.]

>So I suggest to apply the patch to all mipsel boxes.

 

Ok then, for 8 and 8.1 do that.

[WanWizard]

Hmm, when I do that:

ERROR: e2fsprogs-1.45.3-r0 do_patch: Command Error: 'quilt --quiltrc /data/openpli/pli-oe/develop/build/tmp/work/mips32el-oe-linux/e2fsprogs/1.45.3-r0/recipe-sysroot-native/etc/quiltrc push' exited with 0  Output:
Applying patch 005-misc-rename-copy_file_range-to-copy_file_chunk.patch
can't find file to patch at input line 18
Perhaps you used the wrong -p or --strip option?
The text leading up to this was:
--------------------------
|From 01551bdba16ab16512a01affe02ade32c41ede8a Mon Sep 17 00:00:00 2001
|From: Palmer Dabbelt <palmer@dabbelt.com>
|Date: Fri, 29 Dec 2017 10:19:51 -0800
|Subject: [PATCH] misc: rename copy_file_range to copy_file_chunk
|
|As of 2.27, glibc will have a copy_file_range library call to wrap the
|new copy_file_range system call.  This conflicts with the function in
|misc/create_inode.c, which this patch renames _copy_file_range.
|
|Signed-off-by: Palmer Dabbelt <palmer@dabbelt.com>
|Signed-off-by: Theodore Ts'o <tytso@mit.edu>
|---
| misc/create_inode.c | 8 ++++----
| 1 file changed, 4 insertions(+), 4 deletions(-)
|
|--- a/misc/create_inode.c
|+++ b/misc/create_inode.c
--------------------------
No file to patch.  Skipping patch.
4 out of 4 hunks ignored
Patch 005-misc-rename-copy_file_range-to-copy_file_chunk.patch does not apply (enforce with -f)
ERROR: Logfile of failure stored in: /data/openpli/pli-oe/develop/build/tmp/work/mips32el-oe-linux/e2fsprogs/1.45.3-r0/temp/log.do_patch.11729
NOTE: recipe e2fsprogs-1.45.3-r0: task do_patch: Failed
ERROR: Task (/data/openpli/pli-oe/develop/openembedded-core/meta/recipes-devtools/e2fsprogs/e2fsprogs_1.45.3.bb:do_patch) failed with exit code '1'
NOTE: recipe openpli-bootlogo-1.0-r0: task do_install: Succeeded
NOTE: Tasks Summary: Attempted 4303 tasks of which 4297 didn't need to be rerun and 1 failed.

[A.A.]

 

So if you create a e2fspprogrs-dm8000, you have to to that for all packages which depends on e2fsprogs,

I am not sure.

I must study the issue bust most probably it is a single binari provided by e2fsprogs offending the dm8000.

We can mask/replace that binary.

 

At build time all other recipes would DEPEND on the normal e2fsprogs headers.

[Beeker]

@wanwizard.

 

Wrong patch

 

https://github.com/O...f03888deba243ff

[WanWizard]

Sjeesh... I think I'll take the rest of the weekend off...